metadata of the chapter that will be visualized online ... · pdf filewill be at oddswith our...
TRANSCRIPT
Metadata of the chapter that will be visualized online
Chapter Title Does Emergence Also Belong to the Scientific Image? Elements of anAlternative Theoretical Framework Towards an Objective Notion ofEmergence
Copyright Year 2016Copyright Holder Springer International Publishing SwitzerlandCorresponding Author Family Name Huneman
ParticleGiven Name PhilippeSuffixDivision Institut d’Histoire et de Philosophie des
Sciences et des TechniquesOrganization CNRS/Université Paris I SorbonneAddress Paris, FranceEmail [email protected]
Abstract Emergence is a word that plays a central role in the natural or manifestimage of the world, within which we organize our ordinary knowledge.Even though some interpretations of the “scientific image” leave noplace for emergence, sciences increasingly made use of this word. Butmany philosophical arguments have been made against the consistenceor validity of this concept. This chapter presents a computational view ofemergence, alternative to the usual combinatorial view common amongphilosophers, that is formulated in terms of parts and wholes. It showsthat computational emergence can be characterized in terms of causation,and that a subclass of computationally emergent processes displays manyof the connotations of the scientific use of the term. After having socaptured a concept of emergence, I turn to the question of applying theconcept and testing whether some instantiations exist.
UNCORRECTEDPROOF
Chapter 22 1
Does Emergence Also Belong to the Scientific 2
Image? Elements of an Alternative Theoretical 3
Framework Towards an Objective Notion 4
of Emergence 5
Philippe Huneman 6
Abstract Emergence is a word that plays a central role in the natural or manifest 7
image of the world, within which we organize our ordinary knowledge. Even though 8
some interpretations of the “scientific image” leave no place for emergence, sciences 9
increasingly made use of this word. But many philosophical arguments have been 10
made against the consistence or validity of this concept. This chapter presents 11
a computational view of emergence, alternative to the usual combinatorial view 12
common among philosophers, that is formulated in terms of parts and wholes. It 13
shows that computational emergence can be characterized in terms of causation, 14
and that a subclass of computationally emergent processes displays many of the 15
connotations of the scientific use of the term. After having so captured a concept of 16
emergence, I turn to the question of applying the concept and testing whether someAQ1 17
instantiations exist. 18
It is striking that the theory which holds that only entities of fundamental physics 19
are real entities, and therefore claims that predicates like “think”, “believe”, 20
“idea”, “affection”, are illusory – such theory, called “eliminativist” (Churchland 21
1981), seems immediately absurd to most of us. The question therefore raised by 22
eliminativism is whether the ultimate ontology of everything, given by the sciences, 23
will be at odds with our usual knowledge and way of speaking of things – something 24
like the conflict between two “images of the world” as Sellars put it a long time ago – 25
or whether some of the ontological categories proper to our everyday discourses, 26
such as “thoughts”, “cars”, “trees” etc. have ontological relevance. In this latter case, 27
one of the fundamental insights proper to lay knowledge and everyday discourses is 28
P. Huneman (�)Institut d’Histoire et de Philosophie des Sciences et des Techniques, CNRS/Université Paris ISorbonne, Paris, Francee-mail: [email protected]
© Springer International Publishing Switzerland 2016J. Redmond et al. (eds.), Epistemology, Knowledge and the Impact of Interaction,Logic, Epistemology, and the Unity of Science 38, DOI 10.1007/978-3-319-26506-3_22
UNCORRECTEDPROOF
P. Huneman
the idea that some “kinds” of stuff are novel regarding some more “basic” things – 29
in the sense there is something novel in a running cheetah that is not included in the 30
quarks that make it up. 31
The concept of emergence in general aims at making sense of this sense of 32
novelty – of properties, of entities, of laws, etc.1 – within the framework of 33
naturalism, which seems most accurate to the “scientific image” – namely the refusal 34
of dualism, of positing a region of being besides, and independently of, the natural 35
world as unveiled by natural sciences. The idea of emergence therefore rests on 36
the shared intuition that, if, on the one hand, a scientific mind must not admit any 37
supernatural thing, on the other hand an explanation of things such as trends in 38
economy, thought or affects, or history of political ideas, cannot be worked out in 39
terms of motion of quarks or muons, or other elementary entities in particle physics. 40
For these reasons, the word “emergence” is pervasive in the scientific as well as 41
the philosophical recent literatures. Nevertheless, nothing proves that it would resist 42
a rigorous elucidation; it might be the case that such intuition would fade after an 43
attempt to clarifying it. 44
In this chapter, after having reviewed some lay uses of the intuitive notion of 45
emergence in the usual discourse and compared it to scientific uses of the term 46
and philosophical traditional reflections on the concept, I will present what I call 47
a computational concept of emergence, contrasting it with another, more frequent, 48
approach to emergence (called here “combinatorial”). I will show that, on the one 49
hand, it is more satisfying and answers better than the combinatorial concept to 50
some objections raised against the very concept of emergence; and on the other side 51
it includes a causal dimension which makes it into a concept proper to capture what 52
is at stake in many appeals to “emergence” in scientific contexts. The last section of 53
the paper will sketch some applications of this concept to special sciences. 54
22.1 Talking About Emergence: Scientists, Philosophers 55
and Ordinary People 56
To know things we generally start by partitioning them and the world into various 57
kinds. While these partitions are highly culturally dependent, and vary according to 58
the development of a given individual, it is nonetheless a basic fact of knowledge 59
that we organize all our experiences around a partition into kinds, classify things 60
along these partitions, and acquire knowledge in such a framework. Kant thought 61
that this partition – what he calls the “specification of nature into a logical system” 62
(Kant 1987) – was a basic requisite for any knowledge, since we form “empirical 63
1This is a question left open here – it’s enough to point that, following Kim, many philosophicalapproaches of emergence concern the emergence of properties, even if physicists like Laughlin(2005) talk of the emergence of laws. I argued (Huneman 2008b) that one should first of all speakof emergent processes instead of emergence of properties, these ones being emergent only in aderivative way.
UNCORRECTEDPROOF
22 Does Emergence Also Belong to the Scientific Image? Elements. . .
concepts” through comparing and weighting differences and resemblances, and such 64
operation would not be possible if we do not presuppose that things will group into 65
kinds, sub-kinds, and so on. 66
Even if such a partition may vary, in general “living things”, “minds”, “bodies” 67
constitute an important articulation for them, as well as “animals” and “plants”. If 68
there is a natural or “manifest image” of the world, as Sellars (1962) put it, it may 69
clearly include such a subdivision. Cultural anthropology has shown that various 70
cultures will not draw the lines in the same way, many of them having categories in 71
which “life” and “mind” overlap, and a case could be made that it’s mostly western 72
thought from the early modern age on that has insisted on a sharp divide between 73
“minds” or “humans” and living beings (Jonas 1966). Developmental psychologists, 74
on the other hand, after Piaget’s seminal work (1932) accumulated findings about 75
the way western children go through a stage of “animism” where life is a category 76
projected onto all active things, then at age 8–9 restrict this to moving bodies (even 77
falling bodies), and then to bodies that seem endowed with self-motion (e.g. sun, 78
rivers) and finally converge towards an ordinary cultural concept of living things 79
(animal and plants), and then intentional and mental agents. 80
In addition, on this basis many views have been suggested in order to understand 81
how some entities of a given kind can be articulated with entities of another kind: 82
how human beings can have body and mind, how living things can be generated, 83
etc. For instance animism, vitalism (Wolfe and Normandin 2013), and mechanism 84
are families of theories that articulate differently an understanding of what life 85
and livings things are, and how they connect to physical things. Concerning mind 86
and mental states, philosophers have been designing varieties of monism, dualism, 87
panpsychism – even though in many non-Western cultures we fail to see the way we 88
westerners take for granted that dualism has to be taken into account (Descola 2005), 89
so that the “body and mind” problem so familiar to contemporary philosophers of 90
mind does not make sense. 91
Therefore, if we want to roughly sketch the framework for a natural image of the 92
world that is more or less shared by many cultures, and in which people organize 93
their knowledge of things, there is an important room for a notion of how things 94
of one kind may arise on the basis of other extant things. “Emergence”, defined as 95
the “progress of coming into existence or prominence” by the Oxford dictionary, 96
plays this role in scientific and philosophical contexts. It is interesting to see that 97
etymologically it derives from Latin word “emergentia” which means “coming to 98
light”: the ordinary concept, then, carries this connotation that what is emerging was 99
somehow concealed in what it emerges from, or in other word, that what emerges 100
comes from something that had a potential for making it emerging. Many theories 101
that have been elaborated in the past concerning the existence of organisms on the 102
basis of brute matter – “spontaneous generation” – assume that life emerges from 103
dead or brute matter.2 104
2See Roe 1981 on the entanglement of spontaneous generation idea with controversies overgeneration.
UNCORRECTEDPROOF
P. Huneman
Thus ordinary ideas of the nature of main kinds of things in the “natural image” of 105
the world in which our usual knowledge develops include a room for this intuitive 106
idea of emergence. However, a sense of emergence is not at all absent from the 107
modern scientific image: emergence talk indeed occurs precisely when it comes to 108
account for the fact of novelty, or novel kinds of things in a given field. Scientists 109
often use the term “emergence” because many of them believe that even if the 110
investigated phenomena are made of material items which obey laws of elementary 111
physics, such physics is not sufficient to understand them. This is true for human 112
and social sciences but also for biology and even for the regions of physics which 113
are not particle physics (the so-called “fundamental physics”). For philosophers 114
on the other hand, emergentism first means a view defended in the 20 by Samuel 115
Alexander, Lloyd Morgan or C.D. Broad, philosophers who thought that emergence 116
actually reconciled naturalism with the acknowledgment of the existence of novel 117
properties beyond elementary physics. They were proved wrong by the progress of 118
science to the extent that one of their paradigmatic examples was the properties of 119
water – which were, according to them, unexplainable through atomism – and which 120
later have been explained precisely by the quantum physics of covalent linkage (Mc 121
Laughlin 1992). The notion then came back through the field of the philosophy 122
of mind: the main problem here is to understand mental states as, at the same time, 123
grounded upon, and irreducible to, brain states. The discussion then revolved around 124
argument suggested by Jaegwon Kim, who sees mental properties as epiphenomenal 125
ones, because if one is committed to the “causal closure of physics”,3 they cannot 126
have any causal efficiency (all their causal strength comes from their physical bases) 127
and then they have a mere epiphenomenal reality. However, the generality of the use 128
of the word emergence in the sciences contrasts with the specificity of the use of this 129
term by many philosophers. Some of them, considering the concept, come to the 130
conclusion that either there are no emergent properties at all, or only phenomenal 131
consciousness (i.e., “what it’s like” to have this thought or to be this person, e.g. 132
Nagel (1974): “what it’s like to be a bat”) would be emergent (Chalmers 2009).AQ2 133
But, following the general orientation of the volume edited recently by Bedau and 134
Humphreys (2008) I aim here at making sense of the concept of emergence as one 135
can find it in the sciences, instead of discussing what should be the concept of 136
emergence and what would instantiate it in the sole light of the mind/body problem. 137
As said Bedau (2008) a concept of emergence must at the same time mean the 138
autonomy (viz. some bases) and dependency (viz. these bases) of what is emergent. 139
An important distinction has to be made between two different questions regarding 140
emergence, namely the question about the meaning of emergence, and the issue 141
of the reality of emergence. The former is about building a coherent concept of 142
emergence, likely to capture many of the uses of the word in the sciences. The 143
latter is whether there are things in the world that actually fall under this concept. 144
This distinction is necessary, because many arguments in philosophy – first of all 145
by Kim – were directed against the consistence of the concept of emergence, i.e. 146
3Idea that any physical fact or event has a cause which is also physical – notwithstanding whatother facts or causes may exist. This postulate is supposed to be inherent to modern science.
UNCORRECTEDPROOF
22 Does Emergence Also Belong to the Scientific Image? Elements. . .
showing that either it makes no sense or it means some kind of epiphenomenalism. 147
In this perspective, if Kim and his supporters are right what we call “emergent” is 148
not emergent because the very concept of emergence is misconstrued, and therefore 149
the question of checking whether something in the world falls under this concept 150
makes no sense. 151
On the contrary, it is conceivable that we could devise a satisfying concept of 152
emergence, and that in the end nothing empirically falls under this concept – even 153
though in some other possible worlds some possible things may fall under the 154
concept. Construing the concept is the first philosophical question; testing whether 155
what is believed to be emergent and then falling under such concept, actually falls 156
under this concept, and finally whether there exists in the world something which 157
belongs to the extension of such concept, is another question, to be mostly answered 158
by the empirical sciences. Some confusion occurred in the debates because these 159
two questions, relevant to two kinds of investigations, have been conflated. Thus, I 160
mostly here elaborate a concept of emergence, and only the last part of the chapter 161
deals with whether or not something falls under this concept and how we can know 162
it. I start by elaborating a concept of emergence that seems to me valid, and also that 163
is such that satisfying this concept proves to be objective or independent from our 164
cognitive abilities. Then it is shown that, by definition, what satisfies this concept 165
is unpredictable, and then I show that a specification of this concept captures what 166
seems emergent to us in many uses of scientific talk. 167
22.2 Combinatorial and Computational Emergence 168
22.2.1 Characters of Emergence and the Non-triviality 169
Requisite 170
Emergence is often conceived of as the issue of understanding the properties 171
of a whole which would be irreducible to properties of the parts – what I 172
call “combinatorial emergence”. Traffic jams (Nagel and Rasmussen 1994), fads 173
(Tassier 2004), temperature, chromosomes at the time of meiosis, all display a 174
behavior which cannot be understood by adding accounts of the behaviors of their 175
parts. It entails that one sees them as “emergent behavior”: such emergence is often 176
viewed as something proper to the whole and irreducible to the parts. Philosophers 177
like Silberstein (2002), O’Connor (1994), Bechtel and Richardson (1992) tackled 178
the issue of emergence through this scheme of the parts vs. whole. In the same way, 179
Phan and Dessalles (2005) see emergence as a drop in complexity, Wilson (2009) as 180
a decrease in degrees of freedom – contrary to the mere product of properties of parts 181
(where one would find additivity of degrees of complexity/degrees of freedom4). 182
4See Atay and Jost 2004, 18.
UNCORRECTEDPROOF
P. Huneman
But let’s take what is often seen as a famous example of emergence, the 183
segregation model by economist Thomas Schelling (1969): according to their 184
“colour”, agents in an agent-based model will eventually get lumped together in 185
homogeneous clusters, like ghettos in real life. The rules are, as one knows, only 186
to have a slight dislike for being part of the minority (something like “if I am 187
the only green in ten reds, then move”). Besides the important teachings in social 188
sciences of this model (essentially about the limits of a desegregation politics based 189
on education), the behavior “join the group” is not given in the behavioral rules 190
of agents, so it is somehow emerging from the added interactions. But groups are 191
not exactly composed of agents, because those groups subsist even if some agents 192
are added and some emigrate or die” (Gilbert 2002). Therefore, given that parts are 193
transient regarding the whole, a simple view of emergence as irreducibility of the 194
whole to its parts is mistaken. 195
Philosopher William Wimsatt (1997) defined emergence as the “failure of 196
agregativity”. The main issue here is to provide then criteria of agregativity – which 197
means tackling the issue of emergence in an inverse way. Wimsatt’s criteria for 198
failure of agregativity are a sophisticated formulation of what is happening when 199
we say that we cannot reduce the properties of the parts to those of the whole. These 200
criteria are: invariance through substitution of parts; qualitative similarity through 201
addition or subtraction of parts; invariance regarding decomposition-re-aggregation 202
of parts; lack of cooperative/inhibitory interactions. Those are criteria of invariance; 203
thereby they take into account the case of parts which change and alternate in a 204
whole like in the segregation model. However, it seems that, except mass, almost 205
nothing is genuinely aggregative, namely satisfying all these invariance criteria. 206
This is clearly a problem for the combinatorial view. Emergence should surely be 207
ascribed to fewer properties than “everything, except the mass”; therefore it should 208
require an additional criterion which is not provided by such analysis. Here we are 209
left with the idea that emergence comes by degrees.5 However in such view, the 210
meaning of emergence is quite superfluous, it would be enough to talk of degrees of 211
agregativity; the concept of emergence can only have a meaning with this additional 212
criterion according to which emergence is more than a mere lack of agregativity, but 213
precisely it can’t be provided by the combinatorial view. 214
Actually, emergence is supposed to encompass several characteristics: un- 215
predictability, novelty, irreducibility (Klee (1984), Silberstein (2002), O’Connor 216
(1994), Crane (2001), Chalmers (2006), Seager (2005), Humphreys (1997) largelyAQ3 217
concur on these characteristics). Many add “downward causation”, but this is more 218
controversial. Irreducibility understood as irreducibility of properties of the whole 219
to properties of the parts seems now quite trivial given the previous considerations, 220
and too frequent to provide something as “emergence”.6 Concerning novelty, since 221
the properties of the whole are quite always novel regarding properties of the parts – 222
think of colour, or even volume : : : – the real issue is: which novelty should count 223
as emergent? We are left once again with no objective criterion. “Novel” most of 224
5Also Bechtel and Richardson 1992.6See also Bar Yam (2004).
UNCORRECTEDPROOF
22 Does Emergence Also Belong to the Scientific Image? Elements. . .
the times mean what has no name yet in our language (Epstein 1999). Hence this 225
unavoidable characteristics leads to a widely shared conclusion: if emergence has 226
any meaning, it is restricted to epistemological emergence, namely relative to a set 227
of theories and cognitive abilities – perhaps to the exclusion of the exceptional case 228
of qualia (Chalmers 2006; Crane 2001; Seager 2005; O’Connor 1994). Generally, 229
most of the authors oppose epistemological and ontological emergence (the latter 230
being in the real world, the former being defined by the weakness of our analytical 231
or theorizing abilities). Most would conclude that the concept of emergence is 232
undoubtedly epistemological only. A major argument for this conclusion is that, 233
as the example of water for British emergentists can remind it, that what seems now 234
emergent is such only relatively to our theories, and that nothing precludes that a 235
more sophisticated theory could later explain how – to stay in the framework of 236
combinatorial emergence – the properties of the whole result from properties of 237
the parts, or are simply the conditional properties of parts, now actualized. Another 238
argument is the fact that what is real must have causal properties, yet if emergent 239
properties emerge upon some bases, and are not transcendent, they receive their 240
causal powers from those of their bases, so they don’t have any of such powers on 241
their own, and thus don’t have a proper ontological character. Kim’s arguments of 242
exclusion and overdetermination provide the most achieved form of this argument. 243
The rest of this chapter explores another concept of emergence than the 244
combinatorial one; I show that it is immune to the triviality problem revealed 245
by Wimsatt’s non-agregativity criteria, and to the usual verdict that emergence is 246
eventually epistemological, and emergent properties are epiphenomenal. 247
22.2.2 The Incompressibility Criterion and Emergent 248
Processes 249
In the framework of computer simulations one has defined what Bedau (1997) calls 250
“weak emergence”.7 According to the purported criterion, a state in a computationalAQ4 251
process is weakly emergent if there is no shorthand to get to it, except by running 252
the simulation. (“The incompressibility criterion of emergence” – see Huneman 253
2008b; Humphreys 2008; Bedau 2008; Hovda 2008). This approach, amongst the 254
four mentioned connotations of emergence (unpredictability, irreducibility, novelty, 255
downward causation), starts from the notion of unpredictability. 256
Such an approach bypasses the question of the cognitive subjectivity proper to 257
the novelty problem in the former approach, because it’s based on a computational 258
property of algorithmic models. That is why we would have a major clue about 259
emergence which would be, if not ontological, at least objective in the same way 260
as conceptual truths of mathematics are objective, independent of our cognitive 261
capacities or epistemic choices. 262
7Humphreys (1997) is the first systematic investigations of epistemological problems raised by thegeneralized use of simulations in the science. Huneman (2011; 2014a, b) tackled this problem inthe framework of evolutionary explanations.
UNCORRECTEDPROOF
P. Huneman
Yet, one could object that our criterion of incompressibility is only temporary, 263
because we cannot claim that in a remote future, with increased computational 264
capacities, we will be still unable to find analytical shortcuts to reach faster the 265
final state than by simulation. However, here is the sketch of a refutation of such 266
objection. I develop some arguments in favor of the objectivity of computational 267
criteria on the basis of Buss et al. (1992). The basic idea consists in building a set of 268
logical automata whose values change according to a global rule R. Each automaton 269
transforms the value of its cells according to an input 0 or 1. Applying the global 270
rule R depends upon the numbers of each values (q1, q2 : : : ) in the set of automata 271
at step n; the input function which determines then the input of all automata at step 272
n C 1 is determined by R. For this reason the system is perfectly deterministic. 273
For a class of rules, it can be shown that the problem of predicting the state of the 274
automata set at time T arbitrary remote is PSPACE complete (see Box 22.1). This 275
result perfectly illustrates the fact that some computational devices are objectively 276
incompressible. As authors write: “If the prediction problem is PSPACE complete, 277
this would mean essentially that the system is not easily predictable, and that 278
there seems to be no better prediction method other than simulation” (Buss et al. 279
1992, 526) Even with infinite cognitive capacities, there would be a real difference 280
between predictions problems which are PSPACE complete and others, therefore the 281
computational definition of emergence is objective. Weak emergence so defined as 282
inaccessibility except via simulation is then not something trivial since, in this con- 283
text, all global rules which are constant-free are computational in polynomial time, 284
which makes a clear distinction between weakly emergent cases and other ones. 285
Box 22.1: Complexity Classes of Prediction Problems for AutomataInput function:
If Zn D 0, F (n C 1) D g0 (F (n))If Zn D 1, F (n C 1) D g1 (F (n))
Functions g0 and g1 have their values in fq1 : : : : : : .qj : : : : : : .qng.Global rule R: Zi has it s values in f0,1g.Zi D M (Ni (q1) : : : : : : .. Ni (qj) : : : : : : . Ni (qn)) where Ni (qj) is the
number of times the value qj is taken at step i.
Step 0 F1 (0) F2 (0) Fi (0) Fm (0)Step 1 F1 (1) F2 (1) Fi (1) Fm (1)
Step k F1 (k) F2 (k) Fi (k) Fm (k)..Step n F1 (n) F2 (n) Fi (n) Fm (n)
(continued)
UNCORRECTEDPROOF
22 Does Emergence Also Belong to the Scientific Image? Elements. . .
Some global rules are constant-free, meaning that they can be formulatedwith no reference to one of the real values qi : : : ..of the constants, and the othercannot.”If there are as many qi as qj, and for all values of i and j, let Z D 0;otherwise Z D 1” is an example of a constant-free rule. Buss et al. (1992) haveshown that, if the global rule is constant-free, then the problem of predictingthe state of the system at time T is PSPACE-complete; that is why the problemcannot be solved in polynomial time (since we assume that no P D NP andthat NP problems are included in PSPACE problems, so that being PSPACEcomplete implies being a problem such that all other problems can be solvedif such problem can be solved, which makes such a problem at least harderthan NP-complete). A detailed demonstration rests on the fact that constant-free global rules are preserved for any permutation of qi....., which constitutesa major difference concerning the computational pattern of prediction.
22.3 Causation and Computational Emergence 286
The present approach starts with a concept of emergence to show its coherence and 287
plausibility. Another issue is then to decide whether exist some things which, in the 288
real world, fall under this concept, that is, are such that if one has an accurate model 289
of the phenomenon, the model will display properties of computational emergence. 290
It’s conceivable, for now, that there are none, or that we don’t know whether the 291
current models we have, and which speak for the existence of emergent properties, 292
are accurate enough. What is shown until now, is that with incompressibility one has 293
a non-trivial, objective, concept of emergence. Because I am only concerned here 294
with the meaning of emergence and not its actuality, I rely on formal properties of 295
simulations such as cellular automata or genetic algorithms. We can’t rely solely on 296
them to find out instances of the concept of emergence in the world, but here they 297
can allow us to construe and justify a proper concept of emergence. 298
Such concept, starting from the idea of unpredictability, includes the notion of 299
irreducibility. I will now show that the notion of novel order can be included in 300
the intuitive notion of emergence. To this end, I show first that one finds in the 301
computational concept of emergence a dimension of causation, so that it’s not a 302
mere formal notion, whatever the degree to which this concept is instantiated in the 303
real world. From this on, I show (2.3) that this connotation of novel order is likely to 304
be met for a subclass of systems displaying computational emergence. (This section 305
surveys arguments presented in Huneman 2008b). The last section (3) will give 306
some clues for the issue of finding in the real world instantiations of this concept of 307
computational emergence. 308
UNCORRECTEDPROOF
P. Huneman
22.3.1 Causation and Simulations 309
First, this is about answering the objection that starting from the context of simula- 310
tions to conceive of emergence compels one to leave aside the most important thing, 311
namely the fact that one calls “emergent” real processes, which thereby encompass 312
some causation, and possibly raise the issue of downward causation, that is, emer- 313
gent properties of entities (e.g. mental states, fads, standing ovations : : : ) causing 314
backwards effects in their bases (brain states, agents, individual spectators). Peter 315
Corning enunciates such objection very clearly, criticizing in general approaches 316
of emergence relying on computer sciences, such as Holland’s views (Emergence 317
1998), based on a study of genetic algorithms: “Consider Holland’s chess analogy. 318
Rules or laws have no causal efficacy; they do not in fact “generate anything”. They 319
serve merely to describe regularities and consistent relationships in nature. These 320
patterns may be very illuminating and important, but the underlying causal agencies 321
must be separately specified (though often they are not).” (Corning 2002, 26). 322
Yet, simulations actually can include a dimension of causation. Let’s take first 323
cellular automata. A cellular automaton is a set of cells which can be in several 324
possible states, the state of each cell C at time n C 1 being determined by its state 325
at n, through a rule which assigns a state to C at n C 1 according to the state of 326
neighboring cells of C at n. This system is wholly determined. 327
Actually, I argue that there are relations of causation within simulations, which 328
are given by the specifications of properties at successive times in the simulation: 329
some properties of a cellular automaton at time n C 1 have as on the background of 330
the rules a sole cause, namely the properties of it at time n. This argument uses the 331
counterfactualist concept of causation, first elaborated by Lewis (1973) and refined 332
since (e.g. Hall and Paul 2003). According to this concept, A causes B iff “if there 333
had not been A, there would not be B.” (I leave aside some subtle distinctions, 334
which aims at excluding obvious counterexamples from this rough formulation of 335
causation.) 336
Since the rules of the cellular automaton are such that several neighborhoods 337
of the cell C (i; n) yield the same state for C (i; n C 1), one can’t say that “if 338
there had not been the same neighborhood state C (i � j, i C j; n) there would not 339
be C (i, n C 1)”. However there are properties which give rise to a counterfactual 340
dependence between their instantiations at n and at n C 1, as sketched in Box 22.2. 341
Therefore characterizing a class of simulations as computationally emergent should 342
entail a specification of this class in terms of a specific causal pattern. 343
Box 22.2Let’s call A1
n a set of states of cell a1Cmn, m varying from 0 to p (p defined
by the rules of the CA), such that their result at level n C 1 is the state (in theconsidered CA) of a1
nC1. But there are j other sets of states like A1n, such
(continued)
UNCORRECTEDPROOF
22 Does Emergence Also Belong to the Scientific Image? Elements. . .
that their outcome is always a1nC1. We can write the set of these sets of states
A1n, k, k varying from 1 to j. The property P n,1 is then defined as such: a CA
is said to have property P n,1at step n iff it exists k, 0 < k < j C 1, such that it isin a state belonging to a set of states A1
n, k.Now, for any i, and for a given state of ai
nC1 at step n C 1, one can definea set of states Ai
n, j, all of which result into the considered state of ainC1,
and then define the property P n,i. let’s define Q the property of being in thestate fa1
nC1: : : ai
nC1: : : am
nC1g, property instantiated by the CA at step n C 1.Finally we can write that the CA is P at step n iff fit has all the fP n,ig. Then,it is true that: “if the CA had not had property P at n it would have not beenQ at n C 1 – this is a counterfactual dependence, hence a causal relationship”.So, the causal explanation of “having property Q at step n C 1” is “having Pat n”. Thereby there are, in simulations such as CA, causal relations betweensets of states at different steps.
Cellular automaton at step n W a1n; a2
n : : : a1Cpn; a2Cp
n
Property P at Step n A1n;j A2
n;j
Property Q at Step n C 1 W a1nC1; a2
nC1
Causation in CA as counterfactual dependence between properties at somesteps. (After Huneman (2008b)).
22.3.2 Causation and Incompressibility. Emergence as Break 344
Up in Causal Explanation 345
Those counterfactual correlations belong to the set of cellular automata. But when 346
there are emergent properties, this means a specific causal singularity. Quickly 347
said,8 when there is emergence, it means that the causal relationship between two 348
successive states of the system (Ain and Ai
nC1 for all i) can never be traced back to a 349
global law of the system (for example a law for all An). For a cellular automaton, one 350
can go from ajn (with j varying from i C p to i � p, p being defined by the rules of the 351
CA) to ainC1, but not in general in a nomothetic way from An (the set of states aj
n) 352
to AnC1, and even less from any given n to An. This could be a way to make sense, 353
from the viewpoint of computational emergence, of what Wimsatt called failure of
8Demonstration in Huneman 2008b.
UNCORRECTEDPROOF
P. Huneman
agregativity, because any aggregative system is such that we can go in a somewhat 354
continuous way from the local to the global, and write a global law to describe this 355
process. 356
In general, for any system the causal explanations can be of two fashions – 357
either forward on the basis of the elements (forward-local), or backward from the 358
whole (global backward). About a thrown stone, one could write the position of the 359
ball at each instant on a trajectory given by the law of gravitation, or compute the 360
position, instant by instant, according to its prior position. This problem in dynamics 361
is such that both approaches coincide because the trajectory is often integrable. The 362
coincidence between causal explanation means that the step by step explanation, 363
represented by a running cellular automaton, and the explanation by a rule, which 364
represents the jump from the initial state to a step n of the automaton and is given 365
by the motion’s equations, do coincide. When there is computational emergence, 366
we lack such a coincidence: in this sense, the proper character of causation in 367
simulations that represent emergent (in the computational sense) processes, is 368
indeed this fracture within causal explanation. If actually such a coincidence was 369
always by principle available, then we would always have a rule to go from the 370
local (explaining the ainC1 by the (i � k < j < i C k) aj
n) to the global, whereas, as 371
we just saw it, this is not the case for emergence because incompressibility means 372
the lack of a shortcut that would play the role of the global equation allowing to 373
fit the global-backward and the local-forward explanations. The causal signature of 374
computational emergence is thereby the break up between those two modalities of 375
causal explanation. 376
22.3.3 Emergent Causal Reliability and Emergent Order 377
Nevertheless there is, among computationally emergent phenomena, a subclass of 378
processes such that, beyond some step, regularities between sets of cells arise. For 379
example, think of gliders and glider guns in Conway’s Game of Life, which is a two- 380
dimensions cellular automaton whose rules are given in Box 22.3 (Gardner 1970). 381
382
Box 22.3: Rules for the Game of LifeConsider a two-dimensional cellular automaton, in which each step can be intwo states, alive or dead. The transition rules from step n to step n C 1 are:
At each step in time, the following transitions occur:
1. Any live cell with fewer than two live neighbours dies, as if caused byunder-population.
2. Any live cell with two or three live neighbours lives on to the nextgeneration.
(continued)
UNCORRECTEDPROOF
22 Does Emergence Also Belong to the Scientific Image? Elements. . .
3. Any live cell with more than three live neighbours dies, as if byovercrowding.
4. Any dead cell with exactly three live neighbours becomes a live cell, as ifby reproduction.
Gliders and glider guns are common term for recurring patterns that occur within 383
it, and that look like flying gliders thrown from a stable device (Fig. 22.1) Here, we 384
could say it in a counterfactual way: if the set of states (defining a glider or glider 385
gun) had not been there (in this position), the glider (as a set of cell-states) would 386
not be in the state one finds it. 387
These dependencies between partly global states of the simulation are not given 388
with the initial rules, which concern only sets of individual cells. In the usual sense, 389
these emerge in the course of the simulation, and can concern a mere transient 390
state of it. But when they happen, they allow a much more simple explanation of 391
the behavior of the simulation than appealing to the rules that govern each cell’s 392
behavior. Why simple ? Because usually one has to specify the states of all cells 393
in order to step by step explain the simulation, whereas here when a rule (as a 394
transient counterfactual dependency) has emerged, it can be stated by specifying 395
positions of sets of states only. This is a coarse-grained explanation (gliders flying, 396
loop self-replicating in Langton’s loop improved by Sayama9), which of course 397
Fig. 22.1 Gliders in a Game of Life simulation. In this grid, cells are either white or black, andthe state of a is determined by the state of the parent cells (white/black) and its eight neighborsaccording to a rule. Gliders are these patterns of black dots extended through several lines that areconserved as such along many steps of the simulation, therefore that seem to “move” (translatewhile rotating) regularly through the grid towards the bottom right, even though the cellularautomaton only determines the state of cells at each step of the run of the simulation
this
figur
ew
illbe
prin
ted
inb/
w
9Langton 1989. On this loop see Salzberg et al. 2003; Sayama 1998.
UNCORRECTEDPROOF
P. Huneman
Fig. 22.2 Filter (Crutchfieldand Hanson 1993) intendingto reveal “mechanical”entities (greek letters inbottom diagram) whichcausally explain it
omits details, but in the same time saves both computation time and information, 398
and allows reliable generalizations, like Epstein’s civil violence study (see below). 399
Israeli and Goldenfeld (2004) have shown that most of the CA rules support, at 400
some point, to be formulated as coarser grain rules, the sets of cells being then taken 401
as cells, so that apparently incompressible rules can be translated, through a coarse 402
grained description, into compressible rules. (Of course in many cases the coarse 403
grained cell is what indeed emerges, sensu the incompressibility criterion, in the 404
simulation) 405
More technically, Hanson and Crutchfield (1993, 1996, Shalizi and Crutchfield 406
2002) developed a method for reading CA in terms of “mechanics”: they identify 407
patterns (named, by analogy with mechanics: lines, points, particles, etc.) whose 408
correlations as such that they underwrite the running of the whole CA (Fig. 22.2). 409
Here, clearly we get a causal lexicon that enables one to make sense of the 410
intuition that emergent properties are such that they encompass another kind of 411
causation. This novel causality is, first of all, counterfactual regularity between sets 412
of cells (in CAs) or agents (in ABMs10). In such a case, reliable predictions and
10For emergence in ABM according to my criteria, see Wilson 2010.
UNCORRECTEDPROOF
22 Does Emergence Also Belong to the Scientific Image? Elements. . .
descriptions of the system can be given at the level of this novel causality, using 413
such dependencies (hence the term “reliable causal emergence” I used to name this 414
subclass of computational emergent processes). 415
22.4 Applications 416
This reliable emergence – in the sense of the subclass of emergent processes 417
satisfying the clause 3.3. of causal reliability – is instantiated by numerous models 418
in empirical sciences. Take the study of fads (Tassier et al. 2004). In this case, one 419
can see clear relations of causality between states of fads at some moments, which 420
define a general pattern of fads processes. According to the parameters, in these 421
multi-agent models the agents either separate into clusters, any of which adopting a 422
given fashion – or display a behavior such that cycles of fads become visible. 423
In the same way, emergence of local norms (Burke et al. 2006) appears as 424
a computationally incompressible process leading to specific patterns, possibly 425
alternate, or fix. The whole system satisfies the computational emergence criterion 426
because one can’t analytically derive the result. Also, emergence of traffic jams 427
(Nagel and Rasmussen 1994) display patterns whose arising process, modeled by 428
a CA, is incompressible. In the same way, once a traffic jam has appeared, it is 429
likely to show causal relationships between itself and, either other traffic jams, or 430
some state variables of the system such as the average speed of cars etc. Finally 431
lipid membranes (Rasmussen et al. 2002) satisfy the same criteria concerning their 432
creation, the authors describing as a discrepancy between two languages the same 433
difference here described as a distinction between a global rule and the lack of 434
immediate transduction into global rules in the case of emergent processes. 435
With Epstein’s work on civil violence (Epstein 2002) one sees a peculiar kind of 436
counterfactual dependence.11 In these models, Epstein defines agents representing 437
social individuals, and studies their propensity to rebel. Each model implements 438
quite intuitive rules, according to which the acting out (violently) of an agent against 439
the State depends both upon the perceived risk, and the frequency of agents already 440
acting out in her neighborhood. This is a typical multi-agent model. Two parameters 441
are defined: level of oppression and level of legitimacy of the government. By 442
varying these parameters, and multiplying simulations, Epstein can show numerous 443
counterfactual dependencies between the values of parameters (or their variations) 444
and global outcomes of simulations (namely the frequency and generalization (or 445
not) of a rebelling behavior.) Here, dependency does not take place between two 446
moments of the simulation, but – on the basis of a large set of simulations – between 447
range of values of parameters and classes of global outcomes (see also Huneman 448
2012). This latter dependency is a generalization of the former one, hence causation 449
11A more precise description of levels of counterfactual dependency, defining modes of regularityand prediction, is done in Huneman (2012).
UNCORRECTEDPROOF
P. Huneman
(counterfactual) at an upper level. One of the most striking results of this study is 450
that, when the legitimacy of a government drops, this can increase the probability of 451
a violent uprising – however the relevant variable here is not the width of the drop, 452
but its speed: a small but quick drop of legitimacy more easily entails an uprising 453
than a much larger but slower drop. 454
22.4.1 Robustness Analysis 455
A last concern could be the following. I have shown that computational emergence 456
is objective, that it concerns some specific causal explanations, that it allows one 457
to define a subclass of emergent phenomena (called ‘reliable emergence’ here) 458
displaying causal relations such that one can recover the idea – proper to our 459
intuitive conception of emergence – of a novel and spontaneous order. Yet someone 460
could always make the following objection: even if such a notion is formally correct, 461
when one asks what instantiates it in the real world, some phenomena such as 462
the ones described indeed instantiate it only under the condition of our accepting 463
the models which describe them. In other words, if this concept of emergence is 464
ontological, it is however not buffered against the fact that nothing would instantiate 465
it in the real world because all models which make us conclude that it is instantiated 466
can, 1 day, be superseded by models with no emergence. Many arguments have been 467
used against such objection (e.g. Bedau 2008; Humphreys 1997, etc.). I indicated 468
(Huneman 2012) the idea that robustness analysis as done by scientists usually 469
would provide a way to conclude positively regarding the genuinely emerging 470
character of some phenomena, independently of the model. 471
The idea of robustness analysis is the following (Levins 1966; Weisberg 2006). 472
To build a model implies choosing parameters and ascribe values to them; this 473
choice of course often involves simplifying assumptions, namely, betting that some 474
parameters are not so relevant to the phenomenon. Behaviors and general outcomes 475
of the model can vary with those values, and more generally with the choice of 476
parameters themselves. In Epstein’s model of civil uprising for example, education 477
is not a parameter. One could add it, and then check whether the model behaves 478
similarly when education level varies. We say that an agent-based model is robust if 479
its range of qualitative behaviors does not change when the parameters themselves 480
change. For example if education would not change anything to Epstein’s results, 481
this would speak for the robustness of the model (across known parameters). 482
Levins (1966) original paper contrasts a model of the evolution of some trait in 483
a population approached via an analysis of gene frequencies, and another model 484
approached as an optimizing process of the phenotypic values. In his perspective, 485
the former (A) is more realist (since it takes into account the genetic make up 486
of organisms) whereas the latter (B) is more general (since it could be applied to 487
various species that have these traits). His general claim is that models indeed may 488
UNCORRECTEDPROOF
22 Does Emergence Also Belong to the Scientific Image? Elements. . .
privilege either realism, or generality – or even precision, e.g. predictive accuracy – 489
one over the other since all these epistemic values can’t be maximized at the same 490
time. It results that the “theorems” that can be derived in both models, e.g. (A) and 491
(B), are called “robust” theorems, and Levins contends that they have good chances 492
to be verified in the real system under study. Actually both Levins’ claims have 493
been challenged: some authors argued that the idea of being compelled to trade-off 494
epistemic values is overstated (Orzack and Sober 1993), while others argued that 495
“robust theorems” are not, as such, likely to be “true” but rather, they need additional 496
empirical confirmation to be taken as true (Kuorikoski et al. 2012). Yet, this view 497
triggered many reflections, beyond the epistemology of ecology, especially about 498
the kinds of trade-off between epistemic values that characterize different fields or 499
subfields (e.g. Matthewson and Weisberg 2009, on economy). 500
Let’s go back to our examples of agent-based models, and assume that we have a 501
model of a target system with a given set of parameters, which satisfies the two 502
emergence clauses and the reliability clause. And let’s now assume that testing 503
most of the parameters, adding some and changing some of them, there is still the 504
finding that reliable computational emergence occurs in models for some values 505
of the variables. This finding is therefore robust across models and, according to 506
Levins’ general argument, that means that it can somehow corresponds to reality. 507
The phrase “being true”, or “capturing reality”, is as such not very precise, and 508
for the present purpose I take it that it means, among other notions, the idea of 509
“capturing the causal structure of the phenomenon”, as Wimsatt (1987) emphasized. 510
Many arguments then support the idea that such propositions robustly established 511
through a family of models describe the causal structure of the phenomenon: one can 512
appeal to an argument of the type “inference to best explanation” (Lipton 1991) in 513
order to reach such conclusion (i.e. nothing explains the fact that a same proposition 514
is derived across many models embedding a variety of parameters better than the 515
fact that the content of such proposition is true), or one can conceive of an argument 516
stating that causation is based on counterfactual dependencies (Lewis 1973; Hall 517
and Paul 2004), and, since the model displays dependencies between intervening 518
variables and then causality (Woodward 2003), robust findings in the model capture 519
these dependencies. 520
From this on, such set of agent-based models, describing the causal structure 521
of reality through robust findings, and satisfying criteria for emergence, makes 522
it legitimate to say that the process at stake in the system under study presents 523
emergent properties. More precisely, since the signature of emergence is a certain 524
specificity of the causal structure – as seen in the preceding section – it is clear 525
that if a finding in a model captures the causal structure of reality and displays 526
emergence, then the causal structure of reality will carry the signature of emergence 527
and therefore real processes will be emergent (be it reliable emergence or mere 528
emergence). In other words robustness analysis of the models is the last step that 529
allows one to infer in some given case, from the satisfaction of formal emergence 530
criteria to the reality of emergent processes or properties. 531
UNCORRECTEDPROOF
P. Huneman
22.5 Conclusion 532
This chapter explored various aspects of a theory of emergence based on the 533
idea of computational emergence. It starts from the everyday use of the word 534
emergence, which has an important role in the natural image in which we construct 535
our knowledge. It is then argued that in the sciences the concept of emergence is 536
increasingly used to make sense of novelty, which is pervasive in the natural image, 537
and then the question arises of whether this is a consistent and well-formed concept. 538
The philosophical analysis posed here intends to answer this question by showing 539
that there is a rigorous way of forging such a concept of emergence that is likely 540
to do the job it is expected to do in various scientific contexts. To this extent, 541
the sciences do not in principle contradict the daily knowledge (embedded in the 542
natural language) that some things/properties emerge. Emergence may not be the 543
place where the scientific image breaks up with the natural image of the world. 544
However it was shown that the concept of emergence adequate to scientific practice 545
has to be construed in a very specific manner: computational incompressibility is the 546
proper criterion for emergence, because the combinatorial approach suffers from the 547
triviality problem. Especially, the concept of emergence so construed is objective 548
and non-epistemic, in the sense that it does not depend upon cognitive capacities 549
and achievements of the subjects. Nevertheless, such concept is not restricted to 550
characterizing purely formal relations, since it can specify the causal signature 551
proper to emergence as a breaking between both aspects (forward-local, backward- 552
global) of causal explanation. 553
In this perspective, what corresponds to the intuitive notion of emergence found 554
under various guises in the natural image is the subclass of processes that are 555
computationally emergent and that, moreover, were characterized above as reliable 556
causal emergence (i.e., that feature at some point coarse-grained counterfactual 557
dependencies). At least the world of numerical simulations provide numerous 558
examples of that, from the Game of Life trough Langton’s loop to Holland’s Echo 559
(Holland 1995, 1998). 560
Philosophy can’t say much more about emergence: whether this concept is 561
instantiated or not in the real world is an empirical question, to which only science 562
can answer – and, the answer seems to be affirmative, regarding the examples given 563
here. But considering scientific models from the computational perspective allowed 564
us at least to bypass metaphysical objections against the concept of emergence itself. 565
The last philosophical point that can be made concerns an answer to the skeptics’ 566
objection contending that anything assertable about our models is subject to a 567
principled suspicion when applied to the reality that they model. In effect, robustness 568
analysis combined with arguments of the kind “inference to best explanation” 569
is likely to make the computational concept of emergence into a concept that 570
is applicable to state of affairs in the world. From there, the question of what 571
instantiates such concept, and how we can prove that it is indeed instantiated in 572
the world, is just beginning. 573
UNCORRECTEDPROOF
22 Does Emergence Also Belong to the Scientific Image? Elements. . .
References 574
Anderson, P.W.: More is different: broken symmetry and the nature of the hierarchical structure ofAQ5 575science. Science 177, 393–396 (1972) 576
Atay, F., Jost, J.: On the emergence of complex systems on the basis of the coordination of complex 577
behaviors of their elements. Complexity 10(1), 17–22 (2004) 578
Bar Yam, Y.: A mathematical theory of strong emergence using multiscale variety. Complexity 579
9(6), 15–24 (2004) 580
Bechtel, W., Richardson, R.: Emergent phenomena and complex systems. In: Beckermann, A., 581
Flohr, H., Kim, J. (eds.) Emergence or Reduction? pp. 257–287. de Gruyter, Berlin (1992) 582
Bedau, M.: Weak emergence. In: Tomberlin, J. (ed.) Philosophical Perspectives: Mind, Causation, 583
and World, vol. 11, pp. 375–399. Blackwell Publishers, Oxford (1997) 584
Bedau, M.: Is weak emergence just in the mind? Mind. Mach. 18, 443–459 (2008) 585
Bedau, M., Humphreys, P.: Emergence. Contemporary Readings in Philosophy and Science. MIT 586
Press, Cambridge (2008) 587
Burke, M., Furnier, G., Prasad, K.: The emergence of local norms in networks. Complexity 11(5), 588
65–83 (2006) 589
Buss, S., Papadimitriou, C., Tsisiklis, J.: On the predictability of coupled automata: an allegory 590
about chaos. Complex Syst. 5, 525–539 (1992) 591
Chalmers, D.: Strong and weak emergence. In: Clayton, P., Davies, P. (eds.) The Re-emergence of 592
Emergence, pp. 244–256. Oxford University Press, Oxford (2006) 593
Churchland, P.M.: Eliminative materialism and the propositional attitudes. J. Philos. 78, 67–90 594
(1981) 595
Corning, P.: The re-emergence of “emergence”: a venerable concept in search of a theory. 596
Complexity 7(6), 18–30 (2002) 597
Crane, T.: The significance of emergence. In: Gillett, C., Loewer, B. (eds.) Physicalism and Its 598
Discontents, pp. 207–224. Cambridge University Press, Cambridge (2001) 599
Crutchfield, J., Hanson, J.: Turbulent pattern bases for cellular automata. Phys. D 69, 279–301 600
(1993) 601
Crutchfield, J., Hanson, J.: Computational mechanics of cellular automata: an example. Phys. D 602
103, 169–189 (1997) 603
Crutchfield, J., Shalizi, C.: Pattern Discovery and Computational Mechanics. arXiv:cs/0001027v1 604
(2001) 605
Descola, P.: Par-delà nature et culture. Gallimard, Paris (2005) 606
Epstein, J.: Modeling civil violence. Proc. Natl. Acad. Sci. U. S. A. 99(3), 7243–7250 (2002) 607
Epstein, J.: Agent-based computational models and generative social science. In: Generative Social 608
Science: Studies in Agent-Based Computational Modeling, pp. 4–46. Princeton University 609
Press, Princeton ([1999] 2007) 610
Gilbert, N.: Varieties of emergence. Paper presented at the Social Agents: Ecology, Exchange, and 611
Evolution Conference, Chicago. http://www.soc.surrey.ac.uk/staff/ngilbert/ngpub/paper148_ 612
NG.pdf (2002) 613
Hall, N., Paul, D.: Causation and Counterfactuals. MIT Press, Cambridge (2004) 614
Hanson, J., Crutchfield, J.: Computational mechanics of cellular automata: an example. Phys. D 615
103, 169–189 (1997) 616
Hillis, D.: The Pattern on the Stone: The Simple Ideas That Make Computers Work. Basic Books, 617
London (1998) 618
Holland, J.: Hidden Order. How Adaptation Builds Complexity. Addison-Wesley, New York (1995) 619
Holland, J.: Emergence. From Chaos to Order. Basic Books, New York (1998) 620
Hovda, P.: Quantifying weak emergence. Mind. Mach. 18, 461–473 (2008) 621
Humphreys, P.: How properties emerge. Philos. Sci. 64, 53–70 (1997) 622
Humphreys, P.: Synchronic and diachronic emergence. Mind. Mach. 18, 431–442 (2008) 623
Huneman, P.: Combinatorial vs. computational views of emergence: emergence made ontological? 624
Philos. Sci. 75, 595–607 (2008a) 625
UNCORRECTEDPROOF
P. Huneman
Huneman, P.: Emergence and adaptation. Mind. Mach. 18, 493–520 (2008b) 626
Huneman, P.: Computer sciences meet evolutionary biology: issues in gradualism. In: Torres, J.L., 627
Pombo, O., Symons, J., Rahman, S. (eds.) Special Sciences and the Unity of Science, pp. 200– 628
225. Springer, Dordrecht (2011) 629
Huneman, P.: Determinism and predictability: lessons from computational emergence. Synthese 630
185(2), 195–214 (2012) 631
Huneman, P.: Mapping an expanding territory: computer simulations in evolutionary biology. Hist. 632
Philos. Life Sci. 36(1), 60–89 (2014a) 633
Huneman, P.: Mapping an expanding territory: computer simulations in evolutionary biology. Hist.AQ6 634Philos. Life Sci. 34(2) (in press) (2014b) 635
Israeli, N., Goldenfeld, N.: On computational irreducibility and the predictability of complexAQ7 636physical systems. Phys. Rev. Lett. 92, 074105 (2004) 637
Jonas, J.: The Phenomenon of Life: Toward a Philosophical Biology. Harper & Row, New York 638
(1966) 639
Kant, I.: Critique of Judgment. Hackett, Indianapolis (1987 [1790]) 640
Kim, J.: Making sense of emergence. Philos. Stud. 95, 3–36 (1999) 641
Klee, R.: Microdeterminisms and concepts of emergence. Philos. Sci. 51, 44–63 (1984) 642
Kuorikoski, J., Lehtinen, A., Marchionni, C.: Robustness analysis disclaimer: please read the 643
manual before use! Biol. Philos. 27, 891–902 (2012) 644
Langton, C.: Artificial life. In: Langton, C. (ed.) Artificial Life. SFI studies in the sciences of 645
complexity, Proc. Vol. VI. Addison-Wesley, Redwood City (1989) 646
Laughlin, R.: A Different Universe: Reinventing Physics from the Bottom Down. Basic Books,AQ8 647New York (2005) 648
Laughlin, R.B., Pines, D., Schmalian, J., Stojkovi, B., Wolynes, P.: The middle way. Proc. Natl. 649
Acad. Sci. U. S. A. 97(1), 32–37 (2000) 650
Levins, R.: The strategy of model building in population biology. In: Sober, E. (ed.) Conceptual 651
Issues in Evolutionary Biology, 1st edn, pp. 18–27. MIT Press, Cambridge, MA (1966) 652
Lewis, D.: Causation. J. Philos. 70, 556–567 (1973) 653
Lipton, P.: Inference to the Best Explanation. Routledge, London (1991) 654
Matthewson, J., Weisberg, M.: The structure of trade-offs in model building. Synthese 170(1), 655
169–190 (2009) 656
Mc Laughlin, B.: The rise and fall of British emergentism. In: Beckermann, A., Flohr, H., Kim, J. 657
(eds.) Emergence or Reduction? de Gruyter, Berlin (1992) 658
Nagel, K., Rasmussen, K.: Traffic at the edge of chaos. In: Brooks, R. (ed.) Artificial Life IV. MIT 659
Press, Cambridge, MA (1994) 660
Nagel, T.: What is it like to be a Bat? Philos. Rev. 435–450 (1974)AQ9 661Newman, D.: Emergence and strange attractors. Philos. Sci. 63, 245–261 (1996) 662
O’Connor, T.: Emergent properties. Am. Philos. Q. 31, 91–104 (1994) 663
Orzack, S.H., Sober, E.: A critical assessment of Levins’s the strategy of model building in 664
population biology (1966). Q. Rev. Biol. 68, 533–546 (1993) 665
Piaget, J.: La construction du réel chez l’enfant. Delachaux et Niestlé, Paris (1937) 666
Rasmussen, S., Barrett, C.: Elements of a theory of simulation. In: Moran, F., et al. (eds.) AdvancesAQ10 667in Artificial Life: Third European Conference on Artificial Life, pp. 515–529. Springer, 668
Berlin/New York (1995) 669
Rasmussen, S., Baas, N., Mayer, B., Nilsson, M., Olesen, M.: Ansatz for dynamical hierarchies. 670
Artif. Life 7(4), 329–353 (2002) 671
Reynolds, C.: Flocks, herds and schools: a distributed behavioural model. Comput. Graph. 21(4), 672
25–34 (1987) 673
Roe, S.A.: Matter, Life, Generation. Eighteenth-Century Embryology and the Haller-Wolff Debate. 674
Cambridge University Press, Cambridge (1981) 675
Salzberg, C., Antony, A., Sayama, H.: Genetic diversification and adaptation of self-replicators 676
discovered in simple cellular automata. In: Proceedings of the Sixth International Conference 677
on Humans and Computers (HC-2003), pp. 194–199. University of Aizu, Aizuwakamatsu 678
(2003) 679
UNCORRECTEDPROOF
22 Does Emergence Also Belong to the Scientific Image? Elements. . .
Sayama, H.: Spontaneous evolution of self reproducing loops in cellular automata. In: Bar-Yam, 680
Y., Minai, A.A. (eds.) Unifying Themes in Complex Systems Volume II: Proceedings of the 681
Second International Conference on Complex Systems, pp. 363–374. Westview Press (1998) 682
Schelling, T.: Models of segregation. Am. Econ. Rev. 59(2), 488–493 (1969) 683
Seager, W.: Emergence and efficacy. In: Erneling, C., Johnson, D. (eds.) The Mind as a Scientific 684
Object Between Brain and Culture, pp. 176–192. Oxford University Press, Oxford (2005) 685
Sellars, W.: Philosophy and the scientific image of man. In: Colodny, R. (ed.) Frontiers of Science 686
and Philosophy, pp. 35–78. University of Pittsburgh Press, Pittsburgh (1962) 687
Shalizi, C., Haslinger, R., Rouquier, J.B., Klinkner, C., Moore, C.: Automatic filters for the 688
detection of coherent structures in spatiotemporal systems. ArXiv CG/0508001 (2006) 689
Silberstein, M.: Reduction, emergence and explanation. In: Silberstein, M., Machamer, P. (eds.) 690
Blackwell Guide to the Philosophy of Science, pp. 80–107. Blackwell, Oxford (2002) 691
Tassier, T.: A model of fads, fashions and group formations. Complexity 9(5), 51–61 (2004) 692
Weisberg, M.: Robustness analysis. Philos. Sci. 73, 730–742 (2006) 693
Wimsatt, W.: False models as means to truer theories. In: Nitecki, N., Hoffman, A. (eds.) Neutral 694
Models in Biology, pp. 23–55. Oxford University Press, Oxford (1987) 695
Wimsatt, W.: Aggregation: reductive heuristics for finding emergence. Philos. Sci. 64, S372–S384 696
(1997) 697
Wolfe, C., Normandin, S.: Vitalism and the Scientific Image in Post-Enlightenment Life Science, 698
1800–2010. Springer, Dordrecht (2013) 699
Woodward, J.: Making Things Happen. Oxford University Press, New York (2003) 700
UNCORRECTEDPROOF
AUTHOR QUERIES
AQ1. Please provide keywords.AQ2. Chalmers (2009), Phan and Dessalles (2005), Wilson (2009, 2010), Hall
and Paul (2003), Gardner (1970), Hanson and Crutchfield (1993, 1996),Shalizi and Crutchfield (2002), Tassier et al. (2004) are not provided in thereference list. Please provide.
AQ3. The citations Crane (2006), Gilbert (1992), Israeli and Goldenfeld (1996),Rasmussen et al. (1995), Sayama and Salzberg (2003) have been changedto Crane (2001), Gilbert (2002), Israeli and Goldenfeld (2004), Rasmussenet al. (2002), Salzberg et al. (2003) as per the reference list. Please check ifokay.
AQ4. Please fix “a or b” for Huneman (2014) in footnote 7.AQ5. Please provide in-text citations for Anderson (1972), Crutchfield and
Hanson (1997), Crutchfield and Shalizi (2001), Hanson and Crutchfield(1997), Hillis (1998), Huneman (2008a), Kim (1999), Laughlin et al.(2000), Newman (1996), Piaget (1937), Rasmussen and Barrett (1995),Reynolds (1987), Shalizi et al. (2006).
AQ6. Please provide page range for Huneman (2014b).AQ7. Please confirm the inserted volume number and page range for Israeli and
Goldenfeld (2004).AQ8. Please confirm the inserted publisher location for Laughlin (2005).AQ9. Please provide volume number for Nagel (1974).
AQ10. Please confirm the inserted publisher name and location for Rasmussen andBarrett (1995).